innovating nanoscience High-throughput electronic structure theory: - - PowerPoint PPT Presentation

▶
innovating nanoscience high throughput electronic
SMART_READER_LITE
LIVE PREVIEW

innovating nanoscience High-throughput electronic structure theory: - - PowerPoint PPT Presentation

Jul 25, 2023 160 likes •509 views

MaX Conference, Trieste January 2018 innovating nanoscience High-throughput electronic structure theory: are all calculations useful ? Stefano Sanvito (sanvitos@tcd.ie) School of Physics and CRANN, Trinity College Dublin, IRELAND My objectives

slide-1
SLIDE 1

innovating nanoscience High-throughput electronic structure theory: are all calculations useful ?

Stefano Sanvito (sanvitos@tcd.ie)

School of Physics and CRANN, Trinity College Dublin, IRELAND MaX Conference, Trieste January 2018

slide-2
SLIDE 2

My objectives for this talk Demonstrate that HTEST works and that new magnets can be discovered Show that, as databases grow, we will become more clever in creating and using them

slide-3
SLIDE 3

US permanent magnets market ~22.6B$ (2021)

4 Be 9.01 12Mg 24.21 2 He 4.00 10Ne 20.18 24Cr

52.00

19K

38.21

11Na 22.99 3 Li 6.94 37Rb 85.47 55Cs 132.9 38 Sr

87.62

56Ba

137.3

59Pr 140.9 1 H 1.00 5 B 10.81 9 F 19.00 17Cl 35.45 35Br 79.90 21Sc

44.96

22Ti

47.88

23V

50.94

26Fe

55.85

27Co

58.93

28Ni

58.69

29Cu

63.55

30Zn

65.39

31Ga

69.72

14Si

28.09

32Ge

72.61

33As

74.92

34Se

78.96

6 C 12.01 7 N 14.01 15P

30.97

16S

32.07

18Ar 39.95 39 Y

88.91

40 Zr

91.22

41 Nb

92.91

42 Mo

95.94

43 Tc

97.9

44 Ru

101.1

45 Rh

102.4

46 Pd

106.4

47 Ag

107.9

48 Cd

112.4

49 In

114.8

50 Sn

118.7

51 Sb

121.8

52 Te

127.6

53 I

126.9

57La

138.9

72Hf

178.5

73Ta

180.9

74W

183.8

75Re

186.2

76Os

190.2

77Ir

192.2

79Au

197.0

61Pm 145 70Yb 173.0 71Lu 175.0 90Th 232.0 91Pa 231.0 87Fr

223

88Ra

226.0

89Ac

227.0

62Sm 150.4 105 66Dy 162.5 179 85 67Ho 164.9 132 20 68Er 167.3 85 20 58Ce 140.1 13 8 O 16.00 35 65Tb 158.9 229 221 64Gd 157.3 63Eu 152.0 90 66Dy 162.5 179 85 Atomic symbol Atomic Number Atomic weight Antiferromagnetic TN(K) Ferromagnetic TC(K)

Cost Periodic Table

80Hg

200.6

36Kr

83.80

54Xe

83.80

81Tl

204.4

82Pb

207.2

83Bi

209.0

84Po

209

85At

210

86Rn

222

Metal Radioactive Nonmetal BOLD Magnetic atom 25Mn

55.85

96 20Ca

40.08

13Al

26.98

69Tm 168.9 56 312 96

36

78Pt

195.1

1043 1390 629 60Nd 144.2 19 292 < $10/kg $10 - 100/kg $100 - 1000/kg $1000 - 10000/kg >$10000/kg 92U 238.0 93Np 238.0

Finding new magnets: why ?

New tech. to deploy

slide-4
SLIDE 4

~2,000

Magnetism is rare The discover a new useful magnet is a rare event

Fe3O4 SrTcO3

slide-5
SLIDE 5

The magnetic genome project

with Stefano Curtarolo, Duke

slide-6
SLIDE 6

The magnetic genome project

Virtual Materials Growth 1) Simulating existing materials 2) Simulating new materials Rational materials storage Creating searchable database where to store information Materials selection Search the database for 1) new materials, 2) physical insights Robust electronic structure method: density functional theory (VASP) Database Creation (AFLOW) Finding descriptors

slide-7
SLIDE 7

The magnetic genome project

Virtual Materials Growth 1) Simulating existing materials 2) Simulating new materials Rational materials storage Creating searchable database where to store information Materials selection Search the database for 1) new materials, 2) physical insights Robust electronic structure method: density functional theory (VASP) Database Creation (AFLOW) Finding descriptors

Virtual Materials Growth 1) Simulating existing materials 2) Simulating new materials Rational materials storage Creating searchable database where to store information

slide-8
SLIDE 8

The magnetic genome project

ICSD: Inorganic Crystal Structure Database

  • 1,616 crystal structures of the elements
  • 28,354 records for binary compounds
  • 55,436 records for ternary compounds
  • 54,144 records for quarternary and quintenary
  • About 113,000 entries (75.6%) have been assigned a

structure type.

  • There are currently 6,336 structure prototypes.
  • Lots of redundancy

Virtual Materials Growth (existing materials)

Only ~150,000 are known to us

slide-9
SLIDE 9

The magnetic genome project

Duke calculated single elements, binary, ternary and some quaternary (about 100,000) Calculations:

  • AFLOW manages the run (large code)
  • DFT done with VASP (pseudo-potential, plane-wave)
  • Calculations at the DFT GGA-PBE level
  • Relaxation performed à new space group worked out
  • Basic electronic structures collected (including: spin-

polarization, effective mass, magnetic moment, etc.) Virtual Materials Growth (existing materials)

  • S. Curtarolo, W. Setyawan, G. L. W. Hart, M. Jahnatek, R. V. Chepulskii, R. H. Taylor, S. Wang, J. Xue, K.

Yang, O. Levy, M. Mehl, H. T. Stokes, D. O. Demchenko, and D. Morgan, Comp. Mat. Sci. 58, 218 (2012)

slide-10
SLIDE 10

The AFLOW consortium

  • S. Curtarolo, W. Setyawan, S. Wang, J. Xue, K. Yang, R.H. Taylor, L.J. Nelson, G.L.W. Hart, S. Sanvito,
  • M. Buongiorno-Nardelli, N. Mingo, O. Levy, Comp. Mat. Sci. 58, 227 (2012)

www.aflowlib.org

slide-11
SLIDE 11

Heusler alloys ~250 known … ~1000 claimed … ~90 magnetic …

slide-12
SLIDE 12

Heusler alloys ~236,000/0.5M calculated !!

slide-13
SLIDE 13

Database

Rational materials storage

  • S. Curtarolo, W. Setyawan, S. Wang, J. Xue, K. Yang, R.H. Taylor, L.J. Nelson, G.L.W. Hart, S. Sanvito, M.

Buongiorno-Nardelli, N. Mingo, O. Levy, Comp. Mat. Sci. 58, 227 (2012)

www.aflowlib.org

slide-14
SLIDE 14

The magnetic genome project

Virtual Materials Growth 1) Simulating existing materials 2) Simulating new materials Rational materials storage Creating searchable database where to store information Materials selection Search the database for 1) new materials, 2) physical insights Robust electronic structure method: density functional theory (VASP) Database Creation (AFLOW) Finding descriptors

Materials selection Search the database for 1) new materials, 2) physical insights

slide-15
SLIDE 15

Back to the magnets …..

  • S. Sanvito et al., Accelerated discovery of new magnets in the Heusler alloy family, Science

Advances 3, e1602241 (2017)

slide-16
SLIDE 16

A look at the full database Descriptor 0: Enthalpy of formation Energy (Ni2MnAl) < Energy (2Ni + Mn +Al) Property: Can be made ?

Total 235,253 Possible 35,602 Unique 105,212 6,778 Possible Magnetic

slide-17
SLIDE 17

Stability analysis

Descriptor 1: Enthalpy of formation Al Ni Mn

2 Ni + Mn + Al Ni2MnAl

Ni2MnAl

2 Ni + MnAl

MnAl MnNi3 NiAl

1/2 (MnNi3 + NiAl + MnAl)

Ni2MnAl

slide-18
SLIDE 18

Stability analysis

This is very much on-going

(e)$

Ni - Mn - Al

slide-19
SLIDE 19

TM3

Look at the transition metal intermetallics

36,540

slide-20
SLIDE 20

In summary … 36,540 possible à 248 stable 22 magnetic à 8 Robust (∆30 criterion) 236,000 possible à 1550 stable 138 magnetic à 50 Robust

Extrapolating For real …. 70 magnetic predicted 80 magnetic known

slide-21
SLIDE 21

Critical temperature magnetism

Descriptor 2: Critical temperature Known Heusler ferromagnets Co2XY Mn2XY Ni2MnY Rh2MnY Cu2MnY Pd2MnY Au2MnY Fe2MnY Generalized regression model based on valence, volume, spin decomposition Prediction of TC

Material V (Å) µ ΔE (eV) T ….. T Co 47.85 2.0

  • 0.30

3007 352 Mn 48.93 2.0

  • 0.32

3524 760 … … … … … … Mn 54.28 9.03

  • 0.17

1918 ?

slide-22
SLIDE 22

Analysis Co2XY Mn2XY X2MnY

slide-23
SLIDE 23

25 26 27 28 29 30

NV

1 2 3 4 5 6

m (µB)

25 26 27 28 29 30

NV

200 400 600 800 1000 1200

TC (K)

Co2MnTi Co2FeSi Co2AB 1 Co2CrGa Co2MnAl/Co2MnGa Co2NbAl Co2VSn Co2NbSn Co2VZn Co2NbZn Co2TaZn Co2VGa/Co2TiGe Co2VAl Co2AB 2 Co2TiGa Co2TiAl Co2FeSi Co2MnTi Co2MnTi Co2FeGa Co2FeAl Co2MnSi Co2MnGe Co2MnSn Co2MnAl/Co2MnGa Co2CrGa Co2NbAl Co2NbSn Co2CrAl Co2VSn Co2VGa/Co2TiGe Co2VAl Co2TaAl Co2AB 3 Co2VZn Co2NbZn Co2TaZn Co2TaZn Co2TiAl Co2TiGa Co2CrAl

Co2YZ

Slater- Pauling

mX2YZ=NV-24 Co2YZ

slide-24
SLIDE 24

X2MnZ

X2MnZ

4.2 4.3 4.4 4.5

dMn-Mn (A)

200 400 600

TC(K)

NV = 27 = 28 = 29 = 27 = 28 = 29 = 30 = 31 = 32 = 33

4.2 4.3 4.4 4.5

dMn-Mn (A)

1 2 3 4 5

m (µB)

Ru2MnV Pd2MnCu Rh2MnTi Pd2MnZn Pt2MnZn Ru2MnNb Ru2MnTa Rh2MnSc Pd2MnAu Rh2MnHf Rh2MnZr Rh2MnZn

Castelliz- Kanomata curve

slide-25
SLIDE 25

X2MnZ

X2MnZ

  • K. Shirakawa et al., J. Magn. Magn. Mater. 70, 421 (1987)
slide-26
SLIDE 26

Mn2YZ

Mn2YZ

45 50 55 60 65

V (A

3)

  • 500
  • 400
  • 300
  • 200
  • 100

100 200

∆H (meV)

Co2XY Mn2XY

Regular Heusler Inverse Heusler

Mn2CoCr (529) Mn2PtCo (1918) Mn2PtV (3353) Mn2PtPd (3218) Mn2PtRh (3247) Mn2PtGa (2236) Mn2PtIn (841)

slide-27
SLIDE 27

Mn3Ga

  • K. Rode et al., Phys. Rev. B 87, 184429 (2013)

Mn3Ga

slide-28
SLIDE 28

Machine learning workflow

229 candidates 80 used for DFT + ML 149 remaining ML TPR 60% (50:50 population) Don’t calculate 30% = 50 2000 candidates 80 used for DFT + ML 1920 remaining Don’t calculate 30% = ~650 250,000 candidates 1000 used for DFT + ML 249,000 remaining Don’t calculate 30% = ~80,000

slide-29
SLIDE 29

OK, but does all that work?

  • S. Sanvito et al., Accelerated discovery of new magnets in the Heusler alloy family, Science

Advances 3, e1602241 (2017)

slide-30
SLIDE 30

Co2MnTi Courtesy J.M.D. Coey’s Lab (P. Tozman, M. Venkatesan) Prepared by arc melting in an Ar atmosphere

Co2MnTi

TCmeasured = 940K TCpredicted = 938K

slide-31
SLIDE 31

Mn2PtPd

TN1measured = 67K TN1measured = 350K

Courtesy J.M.D. Coey’s Lab (P. Tozman, M. Venkatesan) Complex antiferromagnetic

  • rder
slide-32
SLIDE 32

Bottom line …. Demonstrate that HTEST works and that new magnets can be discovered Show that, as databases grow, we will become more clever in creating and using them Yes, it works!! But a massive effort is needed! Maybe … the algorithm will be clever if the researcher is

slide-33
SLIDE 33

TCD Team: Duke Team:

Tom Archer, Anurag Tiwari, Mario Zic, James Nelson

Stefano Curtarolo, Junkai Xue, Kevin Rasch, Corey Oses